Adaptive Quantization and Fast Error

| Recently, there has been an outburst of research in image and video compression for transmission over noisy channels. Channel matched source quantizer design has gained prominence. Further, the presence of variable length codes in compression standards like the JPEG and the MPEG has made the problem more interesting. Error resilient entropy coding (EREC) 20] has emerged as a new and eeective method to combat catastrophic loss in the received signal due to burst and random errors. In this paper, we propose a new channel matched adaptive quan-tizer for JPEG image compression. A slow, frequency non-selective Rayleigh fading channel model is assumed. The optimal quantizer that matches the human visibility threshold and the channel bit error rate is derived. Further, a new fast error resilient entropy code (FEREC) that exploits the statistics of the JPEG compressed data is proposed. The proposed FEREC algorithm is shown to be almost twice as fast as EREC in encoding the data and hence the error resilience capability is also observed to be signiicantly better. On an average, a 5% decrease in the number of signiicantly corrupted received image blocks is observed with FEREC. Upto a 2 dB improvement in the peak signal to noise ratio of the received image is also achieved.